Prompt engineering basics

Prompt engineering simply means designing prompts on purpose so the AI gives reliable, useful results. Here are the techniques you will use most often.

1. Give examples (few-shot prompting)

If you want a specific style, show it. One or two examples are often more powerful than a long description.

Write 3 product taglines in the same style as these examples:
- "Coffee that wakes up before you do."
- "Shoes that keep up with your plans."

Product: a reusable water bottle.

2. Ask for step-by-step thinking

For problems with several parts (maths, planning, comparing options), ask the AI to work through the steps before giving the final answer. It often makes fewer mistakes this way, and you can check its reasoning.

I earn [amount] per month. My rent is [amount] and my other bills are about [amount].
Work through this step by step, then suggest a simple monthly budget with savings.

3. Set clear limits

Limits help the AI focus. Tell it what to include and what to leave out.

• Word or sentence limits: "under 100 words", "in 3 bullet points"
• Level: "for a complete beginner", "for an expert"
• Exclusions: "no jargon", "do not mention prices"

4. Ask for structured output

When you want to copy the answer into a spreadsheet, document or code, ask for a structure.

Compare ChatGPT, Claude and Gemini for a student. Return a table with columns: tool, good for, free plan (yes/no), one tip.

5. Use delimiters for your text

When you paste text, separate it clearly from your instructions so the AI does not mix them up.

Summarise the article between the triple quotes in 5 bullet points for a busy manager.

"""
[paste article here]
"""

Practice

Take any prompt you wrote in the last lesson and improve it with one technique from this lesson: add an example, a limit, or an output format. Compare the two answers.